I don't know if it's generalizable. Processing raw sensor data requires a knowledge about the dynamic range and knowledge about non linear characteristics of th

I don't know if it's generalizable. Processing raw sensor data requires a knowledge about the dynamic range and knowledge about non linear characteristics of the sensor. Bitdepth isn't the only differentiating factor between raw sensor data. 0x100 doesn't mean the same for 2 different sensors. Also, there is an assumption in the article that the data is linear, but that's not true of all sensors. Some compand their data and there is hardware in the camera to relinearize the sensor output.